The present invention provides a method for identifying a synthetic classifier including contacting at least a first and second samples derived from different groups of a cohort with a first plurality of peptides. The first plurality of peptides includes a first subset of peptides defining at least one naturally occurring amino acid sequence, and a second subset of peptides defining a plurality of variants of the first subset of peptides. The plurality of variants includes, for each one of the first subset of peptides, a variant peptide having at least one of a substitution, a deletion, an insertion, an extension, and a modification. The method further includes selecting at least one of the plurality of variants from the second subset of peptides, and defining a synthetic classifier including the at least one of the plurality of variants that distinguishes between samples derived from the first and second cohorts.